IBM Research India - AI for Business Processes
AI for Business Processes
A business process is a collection of related tasks performed by people or system that accomplish a specific organizational goal. Business processes are the lifeline of enterprises. By standardizing procedures to complete tasks that are important to predefined goals, business processes improve efficiency, ensure accountabilty, streamline inter and intra-organizational communication, and optimize resources.
Disruptions in the execution of processes can significantly impact enterprises and, in some cases, economies, too, as was seen with the disruption of supply chains during the pandemic. Hence, enterprises have increased the focus on identifying and automating as many processes as possible by observing and analyzing the business operations and the underlying IT systems. With the availability of large volumes of business and IT events, metrics, and KPIs, it is possible to use AI-based methods and techniques to forecast and predict deviances and failures in execution, thus proactively preventing them. Furthermore, by analyzing the business and IT operations data, process executions can be optimized by identifying automation opportunities and exploring different automation strategies that include Robotic process automation, and intelligent agents for automation. We cover a broad range of areas dealing with the analysis and automation of processes.
Unified Business and IT Observability
As enterprises accelerate the adoption of digital transformation, the complexity of the IT systems and environment supporting the execution of the business processes continues to increase. New technologies, services, and tools get added to support the execution of business processes. But often, the business and IT operations work in silos. The IT teams have their metrics and key performance indicators (KPI) unrelated to the business KPIs or outcomes. Lack of alignment causes erroneous prioritization of IT tasks leading to delays and inefficiency in executing the process. We want to provide unified observability of business and IT operations. The unified view would allow a business user to answer questions such as “Given an issue with the order intake application, what would be the total order amount and the regions impacted?”, or “How much time do we have to resolve the issue before it impacts the Atlanta region where there is an expected increase in order intake?” This research aims to provide a unified and end-to-end view of the business process and IT services and systems together by using the business observability data and the IT observability data. We are building tools to (i) build the spatial and temporal dependencies between the business and IT metrics, events, and KPIs. (ii) forecast the impact of business or IT events on business KPIs and outcomes, and (iii) build digital twins of business processes and underlying IT systems to create and analyze the impact of unseen business and IT events.
Orchestration of a Hybrid Workforce
As automation becomes more prevalent and intelligent, where bots are being used to do repetitive information acquisition tasks and analysis tasks such as processing a large amount of information, it is becoming increasingly common to consider them as ‘Digital workers’. However, there will always be some decision making tasks that would need humans. Hence, organizations will increasingly have humans, and digital workers collaboratively work together, leading to a hybrid workforce. Given that humans often work in an ad-hoc manner, unlike the programmed digital workers, we are building a platform that supports dynamic and flexible composition and orchestration that identifies the suitable digital workers and humans and sequences them for completion of the task. Additionally, to enable a hybrid workforce to interact with each other, the agents and humans can leverage a natural language interface for their interactions. We are also building tools to analyze the human and agent interactions to identify opportunities for further automation and assess the benefits of automating specific tasks.
AI-powered Business Processes in Supply Chains: Beyond Predictive and Prescriptive Analytics
At the business layer, AI has been successful in predictive (forecasting), prescriptive (next best action), and diagnostic analytics. Currently, the efforts in automated robotic, intelligent, and conversational processes are reaping industry adoptions. However, all of the above, focus on individual business processes. For example, the shipment process in supply chains is primarily concerned with on-time delivery and the current AI engines monitor, analyse, and automate shipment process to improve on-time delivery.
IBM's new generation of AI powered supply chain solutions are ushering unique challenges in business process management. The solutions enhance control towers (real-time analytics dashboards) with actionable intelligence (alerts, reactive resolutions, and proactive prescriptive actions) leading to creation, prioritization, and management of operational business processes. This induces two primary challenges. Firstly, larger business goals would drive individual process lifecycles. For instance, the shipment processes that pursue on-time delivery will now be prioritized based on how they affect a business goal like inventory. Secondly, there would be hundreds of business processes interacting and influencing each other, in the execution of supply chain operations pursing conflicting business goals. At IBM Research, we are developing AI models to address the above two challenges.